Introduction

When the 4m Blanco telescope at CTIO was designed in the late 1960's, it is doubtful that the designers anticipated that it would ever be used at prime focus with any detector other than photographic plates. Wide field imaging was to be done with a camera using a pair of non-achromatic triplet correctors, optimized for use in red and blue light.

The telescope has changed greatly since then, as have other large telescopes constructed during the same era. Imaging is now done almost exclusively with CCDs. Image quality has been significantly improved by careful control of environmental variables and upgrading the optics where feasible (Baldwin et al. (1996)).

A new corrector, the Prime Focus Atmospheric Dispersion Compensator (PFADC) has been installed to take advantage of the telescope's improved imaging capability. The PFADC provides high-quality, wide-field achromatic imaging at prime focus and incorporates atmospheric dispersion compensation (ADC). It is used mainly with a CCD imager and a fiber-fed, multi-object spectrograph known as Argus. Direct photography is still supported, though this option is now little used.

In principle, everything there is to know about a system like this can be computed directly from the optical design. However, there are at least 70 independent variables involved in the design and fabrication of this set of optics, such as spacing, radii, tilts, decenterings and refractive indices. Sufficient error in any one of these is capable of rendering the system's image quality unacceptable.

Each of the parameters can be measured, though always with some uncertainty. The corrector cannot be tested as a unit except on the telescope where the only variables which can be accurately measured are the image size and the optical field angle distortion (OFAD). Photographic plates are the classical and still the most appropriate method of directly measuring the OFAD. The large detector area, flatness, continuous nature of the detecting medium and high dimensional stability of plates makes them ideal for the job. Monolithic CCDs of the requisite size, flatness and number of pixels still lie in the future.

The OFAD coefficients for the old CTIO prime focus UBK-7 triplets were determined experimentally using plates by Cudworth & Rees (1991) and Guo et al. (1993). A similar photographic determination of the OFAD for the PFADC has been made recently by Guo et al. (1996).

Photographic measurements are not sufficient to fully characterize the optics. At CTIO, several instruments with differing optical configurations are used at prime focus and some of the elements of the PFADC are moveable. It is impractical to directly measure the OFAD under all possible permutations. What we have done here is to carefully compare the empirically determined OFAD under a single known set of conditions to the predicted performance under the same conditions and quantify a baseline behavior.

Monte Carlo simulations permit us to show that the observed performance of the optics is within the range which would be expected to occur as a result of normal manufacturing tolerances. This gives us confidence that we understand the corrector and allows us to make useful predictions as to how it can be expected to work in other configurations. The analysis represents a synthesis of theory and measurement and results in a better characterization of the corrector's behavior than would have been possible using the information provided by either computer modeling or direct measurement alone.

The results presented here are intrinsically interesting and not merely to potential users of this corrector. We certainly have benefitted from the exercise. Even the answer to such an apparently mundane question as "What happens when a filter is changed?" can be more interesting and significant than one might think. This kind of sub-arcsecond absolute astrometry will also be necessary for modeling a second ADC corrector now under construction. It will be used with "Hydra-CTIO" (Bardeen 1991), a new multiple object fiber-fed spectrograph now being constructed at NOAO-Tucson.

For the moment, our extrapolations of the PFADC's behavior only involve changes of the optical parameters which deal with the effect of changing filters and the corrector back focal distance. We do not yet have enough information to allow us to do comparisons of direct measurement with theoretical models of the ADC function. This is a different and more complex problem which we hope to study in the near future.